Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Three scenarios for continual learning

About

Standard artificial neural networks suffer from the well-known issue of catastrophic forgetting, making continual or lifelong learning difficult for machine learning. In recent years, numerous methods have been proposed for continual learning, but due to differences in evaluation protocols it is difficult to directly compare their performance. To enable more structured comparisons, we describe three continual learning scenarios based on whether at test time task identity is provided and--in case it is not--whether it must be inferred. Any sequence of well-defined tasks can be performed according to each scenario. Using the split and permuted MNIST task protocols, for each scenario we carry out an extensive comparison of recently proposed continual learning methods. We demonstrate substantial differences between the three scenarios in terms of difficulty and in terms of how efficient different methods are. In particular, when task identity must be inferred (i.e., class incremental learning), we find that regularization-based approaches (e.g., elastic weight consolidation) fail and that replaying representations of previous experiences seems required for solving this scenario.

Gido M. van de Ven, Andreas S. Tolias• 2019

Related benchmarks

TaskDatasetResultRank
Image Classificationpermuted MNIST (pMNIST) (test)
Accuracy97.51
69
Image ClassificationSplit MNIST
Average Accuracy19.96
49
Image ClassificationSplit MNIST S (test)
Task-Averaged Accuracy99.61
18
Reinforcement LearningPusher
Average Returns95
16
Reinforcement LearningBlock Sliding
Retention (Task 1)37
12
Forward TransferPusher Task 3
Forward Transfer (%)109
6
Forward TransferPusher Average over Tasks 2-5
Forward Transfer (%)112
6
Block SlidingBlock Sliding
Task 2 Forward Transfer91
6
Forward TransferPusher Task 4
Forward Transfer Reward (%)105
6
Forward TransferPusher Task 5
Forward Transfer Task Reward110
6
Showing 10 of 14 rows

Other info

Code

Follow for update